Re: [PATCH v0 0/2] virtio-blk and vhost-user-blk cross-device migration

[Michael S. Tsirkin]

On Wed, Oct 06, 2021 at 03:27:59PM +0100, Dr. David Alan Gilbert wrote:
> * Michael S. Tsirkin (mst@redhat.com) wrote:
> > On Wed, Oct 06, 2021 at 02:29:37PM +0100, Dr. David Alan Gilbert wrote:
> > > * Michael S. Tsirkin (mst@redhat.com) wrote:
> > > > On Wed, Oct 06, 2021 at 09:43:52AM +0100, Dr. David Alan Gilbert wrote:
> > > > > * Michael S. Tsirkin (mst@redhat.com) wrote:
> > > > > > On Wed, Oct 06, 2021 at 09:28:50AM +0100, Dr. David Alan Gilbert 
> > > > > > wrote:
> > > > > > > To me it feels the same as the distinction between vhost-kernel 
> > > > > > > and qemu
> > > > > > > backended virtio that we get in net and others - in principal 
> > > > > > > it's just 
> > > > > > > another implementation.
> > > > > > 
> > > > > > In net it's actually like this. Same -device, a different netdev.
> > > > > > 
> > > > > > > A tricky part is guaranteeing the set of visible virtio features 
> > > > > > > between
> > > > > > > implementations; we have that problem when we use vhost-kernel 
> > > > > > > and run
> > > > > > > on a newer/older kernel and gain virtio features; the same will 
> > > > > > > be true
> > > > > > > with vhost-user implementations.
> > > > > > 
> > > > > > That's not new but yes we need to work on this.
> > > > > > 
> > > > > > > But this would make the structure of a vhost-user implementation 
> > > > > > > quite
> > > > > > > different.
> > > > > > > 
> > > > > > > Dave
> > > > > > 
> > > > > > Right. That's why I'm reluctant to just add a new device type that
> > > > > > has special compatibility requirements.
> > > > > 
> > > > > Hmm but there's already another layer of hack^Wabstraction in there 
> > > > > isn't there -
> > > > > there's already:
> > > > >     virtio-blk-pci
> > > > >     virtio-blk-device
> > > > > 
> > > > > created when the user specifies a virtio-blk device?
> > > > > 
> > > > > Dave
> > > > 
> > > > virtio-*-pci is there because it was felt these devices look
> > > > differently from e.g. virtio-ccw so should have a different name.
> > > > virtio-blk-device is an internal thingy, users and guests have no idea.
> > > 
> > > Right, so to do what we're asking here, should we keep the
> > > virtio-blk-pci and instantiate virtio-blk-vhost-user instead of
> > > virtio-blk-device?
> > > 
> > > Dave
> > 
> > I guess that's possible, but we need to pass a bunch of parameters.
> > -drive is probably the right want to do that, right?
> 
> I'm not sure about -drive - isn't that very tied into the block layer?
> 
> Dave


If not, we'll need to add some other object, and tie it's id to the device.

> > > > 
> > > > > 
> > > > > > > > 
> > > > > > > > > > -- 
> > > > > > > > > > MST
> > > > > > > > > > 
> > > > > > > > > -- 
> > > > > > > > > Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
> > > > > > > > 
> > > > > > > -- 
> > > > > > > Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
> > > > > > 
> > > > > -- 
> > > > > Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
> > > > 
> > > -- 
> > > Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
> > 
> -- 
> Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK